Dogs seem to have a lot of variation in their different phenotypes. Is there a limit on the things you could select for in a dog breed? Specifically, could you put selection pressure on a line of dogs to select for Relative IQ, and come up with a super smart dog or would you hit a ceiling?

You could select for intelligence in dogs and you would end up with smarter dogs. I doubt you'd hit a ceiling, but you would need to dedicate many lifetimes to the project to get appreciable results.

One problem that is likely to limit the breeding of an Uber smart dog will be the problems associated with inbreeding. Also, the smarter the dog, the more likely it would be able to escape and breed out, which would be good for the health of the breed, but bad for the super smarts of subsequent generations.

I would like to add just a little bit. There might be a theoretical ceiling to the amelioration of intelligence (for example) by artificial selection pressure within one specie. This ceiling could be reached rather rapidly. As an example, we might extrapolate from a similar experiment which was conducted in Russia, aiming at breeding sociable foxes
( … d_RVDocSum).
In 50 years, dog like foxes were obtained. Strangely enough the sociable foxes turned out to have a lot of common features with dogs (like shorter snout and patched coats).
However, there is no theoretical limit to inteligence improvement by evolution, but I would evaluate that it would take several speciation events to do that. Hence the Uberdog might not have as much in common with a dog anymore. And the many lifetimes it would take would be tens of thousands of years.

Why would speciation events be significant?

On a more practical note what series of tests would one do on a dog to assess intelligence that would then be used as a primary screen over the generations to selective breed for that trait?

On why would speciation events be significant:
Well all this is very theoretical, I was merely pointing out that the genome of current dogs is probably limiting for the acquisition of intelligence. Important changes might be required, especially if the current theory that the development of intelligence and language is based on metaphorical transposition of physical experiences (thus dependant on the body one has) has any truth in it. For example the acquisition of a larynx capable of more refined usage than the current dogs', allowing advanced communication sems like a reasonable requisite to unlimited intelligence acquisition. Therefore, the speciation events required might not be significant but they have a very good probability to be. However if in the imaginary selection process you don't select only intelligence but other dog features, it might take a lot longer than an already extremely long process.
About what primary screen could be used to select for intelligence, We are not talking about a real experiment or something happening or even something planed, it is merely a imaginary projection. I would guess that the more efficient as a first screen would be to select dogs by increasingly sophisticated tests to get their food. I think there are intelligence tests for mice that would do a good start, then the ones for apes, like pushing buttons, the capacity to learn, to mimick, to communicate, etc... It would take ages...

Hi, Jerome, thanks for that.  Let me try to make my question more explicit.  I can see that lot of evolutionary distance would be required to make dogs that are much more intelligent; what I don't see is why speciation is required, rather than a single long-lived and controlled population.  As I understand speciation, we're talking about the reproductive isolation of a subpopulation that was originally part of the main population (and, no I am not trying to get into a discussion the advantages and disadvantages of the various Species Concepts).  What I was trying, clumsily, to ask was this: what role would repeated reproductive isolations play in the artificial evolution of dog intelligence?

Hi, Mike. Sorry, I did not get your question right. Actually if you think about it, artificial selection always involves repeated reproductive isolation or (as Paolo was pointing out in the first post), it has chances to fail. As I understand it, speciation is likely to occur if subgroups of a population are inbred repeatedly over a huge length of time and this is exactly what would be required to select for a super intelligent dog: At each generation you would select for the smartest and breed them together, and theoretically the offspring will be the smartest of the next generation, so you breed them together etc... Leading to reproductive isolation.
To be more specific, I believe the change in the dog genome required for unlimited intelligence acquisition will be very hard to obtain without losing interbreeding capacity (the capacity for the selected individuals to breed with the original population), thus leading to speciation. Such an evolutionary process is thought to involve extended genetic reorganization which should compromize interbreeding. If you allow for the speciation events to occur, you allow for larger changes in genome : chromosome duplication, gene duplication, new genes formed by chromosome translocation.... Duplications in particular could be important as they provide a sort of draft which can evolve more freely because it is not essential for the individual reproduction.

Last edited by Jerome Feldmann (4th Apr 2008 08:26:46)

OK, finally I get it :-)  Thanks for that.  I had thought you were saying that speciation events are a precondition for the development of intelligence, whereas you were actually saying they are an inevitable consequence.

Exactly :-)